Posted on

Amazon Rekognition Introduces Streaming Video Events

Amazon Rekognition Introduces Streaming Video Events

AWS recently announced the general availability of Streaming Video Events, a new feature of Amazon Rekognition to provide real-time alerts on live video streams.

The managed service for image and video analysis can help camera manufacturers and service providers detect objects such as people, animals, and packages in live video streams from connected cameras. Streaming Video Events triggers a notification to the device as soon as the expected object is detected. Prathyusha Cheruku, principal product manager at AWS, explains how it works:

The service starts analyzing the video clip only when a motion event is triggered by the camera. When the desired object is detected, it sends a notification that includes the objects detected, bounding box coordinates, zoomed-in image of the objects detected, and the timestamp. The Amazon Rekognition pre-trained APIs provide high accuracy even in varying lighting conditions, camera angles, and resolutions.

Source: https://aws.amazon.com/rekognition/connected-home

Amazon Rekognition Video relies on Kinesis Video Streams to receive and process the video stream: the AWS::Rekognition::StreamProcessor type creates a stream processor used to detect and recognize faces or to find connected home labels.

To better manage the machine learning inferencing costs, customers can specify the length of the video clips to be processed (between 10 and 120 seconds) and can choose one or more objects such as people, pets, and packages, minimizing false alerts from camera motion events. Cheruku clarifies the benefit of Streaming Video Events over traditional motion detectors:

Many camera manufacturers and security service providers offer home security solutions that include camera doorbells, indoor cameras, outdoor cameras, and value-added notification services to help their users understand what is happening on their property. Cameras with built-in motion detectors are placed at entry or exit points of the home to notify users of any activity in real time, such as “Motion detected in the backyard”. However, motion detectors are noisy, can be set off by innocuous events like wind and rain, creating notification fatigue, and resulting in clunky home automation setup.

According to AWS, service providers can use the feature to create better in-app experiences, for example Alexa announcements such as “a package was detected at the front door”. In a separate article, Mike Ames, Prathyusha Cheruku, and David Robo explain how 3xLOGIC uses the new feature to provide intelligent video analytics on live video streams to monitoring agents.

Streaming Video Events is not the only new feature of Amazon Rekognition. Among the 2022 announcements, Rekognition Video added new languages for text detection, introduced new Face APIs for improved accuracy and improved content moderation.

Video Streaming Events is a feature available in a subset of AWS regions, including Northern Virginia, Ohio, Ireland and Mumbai. The label detection is charged at $0.00817/min, with minute increments. The processing of Kinesis Video Streams is charged separately.

Posted on

Amazon Rekognition Introduces Streaming Video Events

Amazon Rekognition Introduces Streaming Video Events

AWS recently announced the general availability of Streaming Video Events, a new feature of Amazon Rekognition to provide real-time alerts on live video streams.

The managed service for image and video analysis can help camera manufacturers and service providers detect objects such as people, animals, and packages in live video streams from connected cameras. Streaming Video Events triggers a notification to the device as soon as the expected object is detected. Prathyusha Cheruku, principal product manager at AWS, explains how it works:

The service starts analyzing the video clip only when a motion event is triggered by the camera. When the desired object is detected, it sends a notification that includes the objects detected, bounding box coordinates, zoomed-in image of the objects detected, and the timestamp. The Amazon Rekognition pre-trained APIs provide high accuracy even in varying lighting conditions, camera angles, and resolutions.

Source: https://aws.amazon.com/rekognition/connected-home

Amazon Rekognition Video relies on Kinesis Video Streams to receive and process the video stream: the AWS::Rekognition::StreamProcessor type creates a stream processor used to detect and recognize faces or to find connected home labels.

To better manage the machine learning inferencing costs, customers can specify the length of the video clips to be processed (between 10 and 120 seconds) and can choose one or more objects such as people, pets, and packages, minimizing false alerts from camera motion events. Cheruku clarifies the benefit of Streaming Video Events over traditional motion detectors:

Many camera manufacturers and security service providers offer home security solutions that include camera doorbells, indoor cameras, outdoor cameras, and value-added notification services to help their users understand what is happening on their property. Cameras with built-in motion detectors are placed at entry or exit points of the home to notify users of any activity in real time, such as “Motion detected in the backyard”. However, motion detectors are noisy, can be set off by innocuous events like wind and rain, creating notification fatigue, and resulting in clunky home automation setup.

According to AWS, service providers can use the feature to create better in-app experiences, for example Alexa announcements such as “a package was detected at the front door”. In a separate article, Mike Ames, Prathyusha Cheruku, and David Robo explain how 3xLOGIC uses the new feature to provide intelligent video analytics on live video streams to monitoring agents.

Streaming Video Events is not the only new feature of Amazon Rekognition. Among the 2022 announcements, Rekognition Video added new languages for text detection, introduced new Face APIs for improved accuracy and improved content moderation.

Video Streaming Events is a feature available in a subset of AWS regions, including Northern Virginia, Ohio, Ireland and Mumbai. The label detection is charged at $0.00817/min, with minute increments. The processing of Kinesis Video Streams is charged separately.

Posted on

Fauna Transactional Database Introduces Event Streaming

Fauna Transactional Database Introduces Event Streaming

Fauna, the company behind the Fauna transactional database, recently announced the general availability of event streaming, a push-based stream that sends changes at both the document and collection levels to subscribed clients.

Shashank Golla, senior product marketing manager at Fauna, explains:

Fauna’s event streaming employs an open, push-based streaming method to automatically stream real-time data updates to your clients when there is a change in your database. Unlike polling, in event streaming the subscription from the client side happens once and changes are automatically broadcast to the client whenever the subscribed document or collection is updated.

Source: https://fauna.com/blog/event-streaming#ensure-clients-have-least-privilege-access-with-abac

Fauna supports two types of event streaming: document streaming, where the client subscribes to a document reference, and set streaming, where the client subscribes to a set reference and when one or more documents enter or leave the set an event notification is triggered.

A distributed database, Fauna is an object-relational, globally replicated service that supports an indexed-document data model and distributed ACID transactions. A subscription is a connection to the cloud service that is held open by the client through the Fauna driver and set and document streaming features are available using the C#, Go, JavaScript, JVM (Java, Scala) and Python drivers. Explaining how to integrate event streaming using a sample react application, Shadid Haque, developer advocate at Fauna, suggests:

Avoid running a query to fetch a document and then establishing a stream. Multiple events may have modified the document prior to stream startup, which can lead to an inaccurate representation of the document data in your application.

Event streaming databases have become popular in the last few years and the major cloud providers offer different managed options to stream data, including DynamoDB Streams and AWS Kinesis Datastream, Datastream on Google Cloud and Azure Event Hubs. CockroachDB and Astra DB support event-driven architectures using Change Data Capture (CDC). Jeremy Daly, GM of serverless cloud at Serverless Inc, comments in his latest newsletter:

If you are a database provider and you’re not drifting into the world of event-driven architecture, you might as well start looking for something else to do.

To ensure that clients have least privilege access, Golia suggests using ABAC, a Fauna’s extension of the traditional role-based access control:

With ABAC, you can implement least privilege access using streaming and provide real-time changes to only the users who should be receiving the updates.

The following limitations apply to Fauna event streaming: GraphQL subscriptions are currently not supported, a browser can open a maximum of 100 streams and a document stream reports only events for the fields and values within the document’s data field.

Event streaming is charged according to usage and is available in all Fauna pricing plans. Each streamed event counts two read operations and includes 4k bytes read from storage, plus one read operation per additional 4k bytes, per subscriber. One compute operation per subscriber is counted for every second a stream is held open.